2.1. Rationale Lung cancer is the leading cause of cancer mortality In the United States, resulting in 162,000 deaths each year, including 17,000 deaths among African Americans (a tenn we use interchangeably with blacks in this proposal). National statistics on cancer mortality indicate that African American men (but not women) and less educated patients have markedly increased risks of death from lung cancer.^ In contrast, for Hispanics and Asian Americans, mortality rates from lung cancer are comparable to or better than for whites. While prevention of lung cancer is a vitally important issue for public health, it is also crucial to understand systematic variations in processes of care (e.g., continuity of care, provision of appropriate curative, life-extending, or palliative care) and outcomes (e.g., stage at diagnosis, health-related quality of life, and survival) for diagnosed patients. These variations include disparities by race and socioeconomic status (SES) and the mediators of these disparities. Understanding such variations is essential to developing interventions within health care systems and communities to reduce these disparities. This project is Informed by a conceptual model of disparities in health care and health outcomes that postulates they arise from a multifactorial mix of social and clinical factors related to patients, communities, providers and health care systems. Detailed data sources and hierarchical analytic methods can help to distinguish between and among the root causes of disparities, thereby informing the design of interventions to eliminate these disparities more effectively. A key issue for health policy is the extent to which racial and socioeconomic disparities in cancer care processes and outcomes reflect differences in the quality of diagnostic and treatment services between hospitals in which disadvantaged patients are treated and those hospitals serving more advantaged patients.^ It is similariy important to detennine the role of differences in processes of care and outcomes between communities with concentrations of disadvantaged patients and more advantaged communities. If between-hospital or between-community differences are a major component of cancer disparities, then policy solutions can focus on improving care in hospitals and communities with large numbers of minority or low SES patients or on expanding access for patients in these communities to better hospitals in other areas. Alternately, if racial and socioeconomic disparities are mainly evident within hospitals and communities, then attention can be focused on identifying reasons why the health care system provides less effective care to some groups, perhaps for reasons both internal and external to the health care system. These ways in which disparities arise may also vary for cancers with very different patterns of diagnosis, treatment, and survival, such as lung cancer and colorectal cancer. Although both cancers are curable when detected at an early stage, lung cancer lacks effective screening tests and is thus typically diagnosed at an advanced stage, allowing limited opportunities for effective treatment when detected late and leading to relatively high mortality rates as a result. In contrast, colorectal cancer can be detected with effective screening tests that are now used by over half of eligible adults, so it is typically detected at an eariier, more curable stage than lung cancer. However, black and less affluent patients continue to have worse outcomes that are related to more advanced stage and worse survival within stage groupings. Thus, colorectal cancer, in addition to being important in its own right, also offers a model for how outcomes of lung cancer may evolve if more effective screening modalities and/or treatment options become available. From 2003 through 2005, the NCI-funded Cancer Care Outcomes Research and Surveillance (CanCORS) Consortium enrolled approximately 10,000 newly diagnosed lung or colorectal cancer patients in multiple regions and health care systems. Extensive primary data have been collected from surveys of these patients and their family members, surveys of their physicians and surgeons, and extensive reviews of their medical records. Health-related quality of life and survival measures have been ascertained through 15 months after diagnosis, and the Consortium has recently been funded to collect additional data on treatment and outcomes of the cohort through 2010 up to seven years after diagnosis. Of particular relevance to the current proposal, and for the approximately 4000 subjects enrolled in CanCORS from Los Angeles (LA) County and eight counties of Northern California: California Cancer Registry (CCR) data have been obtained in April 2009; Medicare enrollment and claims data for elderly members of this cohort are currently being linked for the years 2002 through 2006 with Centers for Disease Control funding; Medicaid (i.e., Medi-Cal) claims data have been linked to the cohort with support from an ongoing ROI grant funded by the National Cancer Institute. In addition, data from the CCR have recently been obtained for 70,000 state residents diagnosed with lung or colorectal cancer from 2002 through 2004, and Medicare claims are soon to be obtained for approximately 35,000 of these residents who were age 65 or older when diagnosed with lung or colorectal cancer in California. About half of these two larger groups are residents of LA County and the Northern California counties that were Included in CanCORS. Together these data from CanCORS, the CCR, Medicare and Medi-Cal provide an unparalleled resource for studying racial, ethnic, and socioeconomic differences in the outcomes of patients with lung cancer, and for contrasting these differences with comparable data for patients with colorectal cancer. These special resources for comparative analyses of lung cancer and colorectal cancer will enable us to compare and contrast the ways in which race and SES affect outcomes for the two most common causes of cancer mortality in the United States. Building on CanCORS data along with cancer registry. Medicare, and Medi-Cal data from California, Project 3 will address three specific aims and related hypotheses.